backlog(B-0095): escrow rules + naming-collision (vendor sense) + deferred-research migration (Aaron 2026-04-29)

docs/backlog/P2/B-0095-escrow-rules-naming-collision-and-deferred-research-migration-aaron-2026-04-29.md

@@ -0,0 +1,202 @@
+---
+id: B-0095
+priority: P2
+status: open
+title: Escrow — define rules + resolve naming collision with software-vendoring sense + migrate other deferred research from backlog/research/
+tier: factory-hygiene
+effort: M
+ask: Aaron 2026-04-29 — *"backlog add other stuff we need from backlog research to escrow, we also needs rules for what goes in here, shit that's not ready yet, we've also overloaded escrow for this use and the use in software engineering for having copies of all our dependies local native incase the remote dependence disappears kind of like vendoring from the old go days (not exactly, there are modern software escrow too)  backlog"*
+created: 2026-04-29
+last_updated: 2026-04-29
+composes_with:
+  - B-0094
+  - PR-714
+tags: [aaron-2026-04-29, escrow, naming-collision, factory-hygiene, deferred-research, vocabulary, vendoring]
+---
+
+# B-0095 — Escrow rules + naming-collision resolution + deferred-research migration
+
+## Source
+
+Aaron 2026-04-29 input after the B-0094 escrow PR landed:
+
+> *"backlog add other stuff we need from backlog research to
+> escrow, we also needs rules for what goes in here, shit
+> that's not ready yet, we've also overloaded escrow for this
+> use and the use in software engineering for having copies of
+> all our dependies local native incase the remote dependence
+> disappears kind of like vendoring from the old go days (not
+> exactly, there are modern software escrow too)  backlog"*
+
+Three distinct sub-asks wrapped in one input.
+
+## Why P2
+
+The escrow primitive landed in B-0094 / PR #714 (research-deferred
+thesis preservation). It will be reused. Without (a) clear rules
+for entry, (b) naming-collision resolution, and (c) population
+from existing deferred research, the primitive risks becoming
+either an inconsistent grab-bag OR confusing maintainers reading
+the directory.
+
+## Sub-ask 1: Naming collision
+
+The word "escrow" is overloaded between two distinct senses:
+
+1. **Research-grade preservation escrow** (the B-0094 sense
+   landing today). Bounded preservation of a not-yet-ready
+   thesis until it earns operational landing through a
+   defined gate.
+2. **Software-engineering escrow / vendoring** (Aaron's
+   reference). Local copies of dependencies kept native to
+   the project so the project survives the remote
+   disappearing. Modern software escrow + Go's old `vendor/`
+   directory are both in this family.
+
+Aaron flagged the overload. Resolution required.
+
+Options:
+
+- **Option A: Rename the research preservation directory.**
+  e.g. `docs/research/deferred/` or `docs/research/parking-lot/`
+  or `docs/research/reservoir/` — pick a term less collision-
+  prone with software-engineering vocabulary.
+- **Option B: Disambiguate via prefix.** Keep `docs/research/
+  escrowed/` but add a top-level rule that this directory is
+  for research-grade preservation, distinct from the software-
+  engineering-vendoring sense which lives elsewhere (e.g.
+  `references/upstreams/` is already the vendoring home per
+  GOVERNANCE §23).
+- **Option C: Name both senses explicitly** in a glossary
+  entry — research-escrow vs vendor-escrow vs dependency-
+  escrow — and let the directory name stay if the glossary
+  carries the disambiguation.
+
+Recommend running through `naming-expert` to pick the best
+option. Aaron's framing suggests at least surface the
+disambiguation — even if directory name stays, the docs
+need to acknowledge the overload.
+
+## Sub-ask 2: Define rules for what goes in escrow
+
+The B-0094 row defined escrow shape for the specific Aurora
+flywheel thesis. The general rule is implicit. Make it
+explicit.
+
+Candidate rules (to be ratified through ADR or skill-creator):
+
+1. **Eligibility**: substrate is escrow-eligible when (a) it
+   is conceptually substantive enough to warrant preservation,
+   (b) it has a defined gate that would unblock it, and (c)
+   landing it as active substrate now would cause a known
+   failure mode (synthesis-treadmill, premature canonization,
+   carrier-laundering, etc.).
+2. **Required schema**: every escrow file has
+   - §33 archive header (Scope/Attribution/Operational status =
+     research-grade/Lifecycle status = escrowed/Non-fusion
+     disclaimer)
+   - Status header block at top (Status / Gate / Reopen
+     condition / Multi-AI loop policy / Expiration / Created /
+     Last surfaced)
+   - Falsifier gate explicitly defined
+   - Expiration trigger (X substantive rounds) named
+   - Non-activation rule (what the file does NOT authorize)
+3. **Single canonical home**: `docs/research/escrowed/`
+   (subject to sub-ask 1 outcome).
+4. **No-multi-reviewer-loop default**: while escrowed,
+   substrate may not be sent through multi-AI synthesis loops.
+5. **Bilateral clarification carve-out**: Aaron + agent may
+   discuss escrowed substrate.
+6. **Implementation work allowed**: prototype / falsifier-
+   gate work is allowed during escrow.
+7. **Explicit decision required at expiration**: no auto-
+   extend; surfaced escrow demands one of four decisions
+   (extend / run prototype / revise / retire).
+
+These rules already live implicitly in B-0094's body. Promote
+to a `docs/research/escrowed/README.md` (or equivalent contract
+doc) as a real rule-set future agents can cite.
+
+## Sub-ask 3: Migrate deferred research
+
+Aaron asks for "other stuff we need from backlog research to
+escrow" — items currently filed elsewhere that match the
+escrow shape.
+
+Audit candidates (NOT yet exhaustive — survey work pending):
+
+- `docs/backlog/P2/B-0089-veridicality-rainbow-table-canonicalization-research-and-graduation-aaron-ani-amara-2026-04-28.md`
+  — graduation roadmap with a falsifier gate. Possibly
+  escrow-eligible if its work is "not ready yet."
+- `docs/backlog/P2/B-0090-cadenced-lost-substrate-recovery-audit-aaron-2026-04-28.md`
+  — operational rather than research; probably stays in
+  backlog/.
+- `docs/research/aurora-immune-governance-bridge-minimal-2026-04-28.md`
+  — already landed as active research; NOT escrow-eligible
+  (it's the prototype gate for B-0094).
+- Various Amara ferry-pending-absorb rows (if any are still
+  open with un-absorbed research that has a clear gate).
+
+Audit work: read every open backlog row in
+`docs/backlog/P*/` + every research note in
+`docs/research/` not under `escrowed/`, and classify:
+ESCROWABLE / STAYS-IN-BACKLOG / STAYS-AS-ACTIVE-RESEARCH /
+RETIRE.
+
+## Acceptance
+
+- [ ] Sub-ask 1: naming collision resolved (option A / B / C
+      chosen + applied; `naming-expert` consulted).
+- [ ] Sub-ask 2: `docs/research/escrowed/README.md` (or
+      equivalent rule-set doc) lands with the 7 candidate
+      rules above (refined via skill-creator or ADR).
+- [ ] Sub-ask 3: audit of existing deferred-research items
+      complete; eligible items migrated to escrow with full
+      schema; ineligible items retain their current home with
+      rationale documented.
+
+## Why M effort
+
+Each sub-ask is S-effort individually (a glossary entry / one
+small README / per-item migrations). Together they're an M.
+Could split into three rows; consolidating because Aaron's
+input was one wrapper.
+
+## Composes with
+
+- **B-0094** — the worked example escrow. Schema this row
+  formalizes is what B-0094 implicitly defined.
+- **PR #714** — the file currently at
+  `docs/research/escrowed/aurora-autonomous-flywheel-thesis-2026-04-28.md`.
+  Naming-collision resolution may rename the directory; if so,
+  the file moves with it.
+- **GOVERNANCE.md §33** — archive-header schema the escrow
+  contract must compose with.
+- **GOVERNANCE.md §23** — `references/upstreams/` is the
+  vendoring home; helps disambiguate sub-ask 1.
+
+## What this row does NOT authorize
+
+- Does NOT authorize moving the existing B-0094 escrow file
+  before sub-ask 1 (naming-collision) is resolved. Until then,
+  the file stays at `docs/research/escrowed/aurora-autonomous-
+  flywheel-thesis-2026-04-28.md` per the in-flight PR #714.
+- Does NOT authorize bulk-migrating backlog rows to escrow
+  before sub-ask 2 (rules) lands. Without rules, migrations
+  are arbitrary.
+- Does NOT authorize creating new escrow content beyond the
+  in-flight B-0094 work.
+
+## Pickup for future Otto
+
+If picking up this row:
+
+1. Run `naming-expert` on sub-ask 1; surface options to Aaron
+   for decision.
+2. Draft `docs/research/escrowed/README.md` with the 7
+   candidate rules; route through skill-creator workflow OR
+   land as ADR + small README.
+3. Audit deferred-research items per sub-ask 3 surface; file
+   migration PRs one at a time.
+
+Each step is bounded; do not stack.